Ahsan Raza Khan

Human Activity Recognition (HAR) using Channel State Information (CSI) enables energy-efficient and non-invasive healthcare monitoring. However, conventional HAR systems rely on centralised model training, which requires the sharing of raw data, leading to privacy risks, excessive bandwidth usage, and high communication latency that limit scalability. This paper proposes FedFusionQuant (FFQ), a federated learning (FL) framework that jointly performs feature fusion, adaptive aggregation, and quantisation-aware compression during training. A novel federated distance (FedDist) mechanism dynamically adjusts parameter updates using neuron dissimilarity metrics, enhancing generalisation across heterogeneous clients. Meanwhile, quantisation-aware training (QAT) reduces model size and transmission cost while preserving accuracy. Extensive experiments on real CSI data from 30 participants demonstrate that FFQ improves multi-class HAR accuracy by 4.29% and binary fall detection by 5.55% compared to raw fusion models. Furthermore, model compression with QAT achieves a 47% reduction in communication overhead while maintaining accuracy comparable to state-of-the-art methods.

The integration of Variable Renewable Energy (VRE) sources in power systems is increased for a sustainable environment. However, due to the intermittent nature of VRE sources, formulating efficient economic dispatching strategies becomes challenging. This systematic review aims to elucidate the economic value creation of Artificial Intelligence (AI) in supporting the integration of VRE sources into power systems by reviewing the role of AI in mitigating costs related to balancing, profile, and grid with a focus on its applications for generation and demand forecasting, market design, demand response, storage solutions, power quality enhancement, and predictive maintenance. The proposed study evaluates the AI potential in economic efficiency and operational reliability improvement by analyzing the use cases with various Renewable Energy Resources (RERs), including wind, solar, geothermal, hydro, ocean, bioenergy, hydrogen, and hybrid systems. Furthermore, the study also highlights the development and limitations of AI-driven approaches in renewable energy sector. The findings of this review aim to highlight AI’s critical role in optimizing VRE integration, ultimately informing policymakers, researchers, and industry stakeholders about the potential of AI for an economically sustainable and resilient energy infrastructure.

Presents corrections to the paper, Corrections to “A Hybrid Approach for Forecasting Occupancy of Building’s Multiple Space Type”

This systematic review aims to elucidate the economic value creation of Artificial Intelligence (AI) in supporting the integration of Variable Renewable Energy (VRE) sources into power systems. Addressing the economic dispatch challenges associated with integrated power system has become paramount due to the increasing penetration of VRE. This paper reviews the role of AI in mitigating costs related to balancing, profile, and grid with a focus on its applications for generation and demand forecasting, market design, demand response, storage solutions, power quality enhancement, and predictive maintenance. By analyzing the use cases across various renewable energy resources (RERs), including wind, solar, geothermal, hydro, ocean, bioenergy, hydrogen, and hybrid systems, this study highlights AI’s potential to enhance economic efficiency and operational reliability. This review spans the literature from 2014 to 2024, offering insights into the advancements and limitations of AI-driven approaches in the renewable energy sector. The findings underscore AI’s critical role in optimizing VRE integration, ultimately facilitating a more resilient and economically sustainable energy landscape.

Predicting signal blockages in millimetre-wave and terahertz networks is essential for enabling proactive handover (PHO) and ensuring seamless connectivity. Existing approaches utilising deep learning, multi-modal vision and wireless sensing data primarily depend on centralised model training. Although these techniques are effective, they come with high communication costs, inefficient bandwidth usage, and latency issues, which restrict their real-time applicability. This paper proposes a Semantic-Aware Federated Blockage Prediction (SFBP) framework, leveraging the lightweight computer vision technique MobileNetV3 for edge-based semantic extraction, lowering communication and computation costs. Furthermore, we introduce a Similarity-Driven Federated Averaging (SD-FedAVG) mechanism to enhance the robustness of the model aggregation process, effectively mitigating the impact of noisy updates and adversarial attacks. Our proposed SFBP framework achieves 97.1% blockage prediction accuracy, closely matching centralised learning methods, while reducing communication costs by 88.75% compared to centralised learning and by 57.87% compared to FL without semantic extraction. Moreover, on-device inference reduces the latency by 23% compared to centralised learning and 18% compared to FL without semantic extraction, improving real-time decision-making for PHO. Additionally, the SD-FedAVG mechanism improves prediction accuracy under noisy conditions, directly impacting the PHO by reducing the handover failure rate by 7%.

Short-term load forecasting (STLF) is essential for efficient power system operations, including unit commitment and energy market operations. Traditional deep learning models like Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks have shown promise but often struggle with capturing complex temporal dependencies and nonlinearities in electricity load data. This paper introduces SpikeNet, a novel neural network architecture incorporating integrate-and-fire (IF) spiking neurons with Recurrent Neural Network (RNN) and LSTM layers to improve temporal dynamics modelling in STLF. Spiking neurons process information through discrete events "spikes", offering enhanced efficiency and the ability to model temporal patterns more effectively. We evaluate SpikeNet's performance against state-of-the-art models on three diverse datasets: Islamabad Electric Supply Company (IESCO), the IEEE data portal, and Panama City. Our results demonstrate that SpikeNet outperforms traditional models, achieving up to 33% improvement in mean absolute percentage error (MAPE) and up to 42% reduction in inference time. These findings highlight SpikeNet's potential as a more accurate and efficient solution for STLF, with significant implications for real-world power system applications.

Human Activity Recognition (HAR) using channel state information (CSI) is crucial for advancing healthcare applications, enabling non-invasive monitoring for improved patient care. However, traditional HAR techniques mainly rely on centralised data processing, which necessitates the sharing of raw data. This approach raises privacy concerns, leading to inefficient bandwidth usage and incurs significant communication overhead and latency, thereby limiting real-time performance and scalability. This paper introduces FedFusionQuant (FFQ), a novel federated learning (FL) framework that integrates advanced signal processing, feature fusion, and model compression during the training process. The proposed framework utilises the federated distance (FedDist) algorithm to adapt parameter adjustments based on neuron dissimilarity measures, effectively mitigating overfitting. Additionally, the incorporation of quantisation-aware training (QAT) allows the model to maintain high accuracy while substantially reducing model size. Extensive empirical evaluations demonstrate the efficacy of FFQ, showing a significant improvement in classification accuracy and energy efficiency through feature fusion and QAT. Furthermore, model compression with QAT achieves a 47% reduction in communication overhead while maintaining accuracy comparable to state-of-the-art methods.

The occupancy datasets are useful for planning important buildings’ related tasks such as optimal design, space utilization, energy management, maintenance, etc. Researchers are currently working on two key issues in building management systems. First, feasible and economical deployment of indoor and outdoor weather and energy monitoring sensors for data acquisition. Second, the development and implementation of cost-effective data-driven models with regular monitoring to ensure satisfactory performance for occupancy prediction. In this context, we present an occupancy forecasting model for different types of rooms in an academic building. A comprehensive dataset comprising indoor and outdoor environmental variables such as energy consumption, Heating, Ventilation, and Air Conditioning (HVAC) operational details and information on Wi-Fi-connected devices of a campus building, is used for occupants’ count prediction. A Light Gradient Boost Machine (LGBM) is applied for the selection of suitable features. After the feature selection, Machine Learning (ML) models such as Extreme Gradient Boosting (XgBoost), Adaptive Boosting (AdaBoost), Long Short-Term Memory (LSTM) and Categorical Boosting (CatBoost) are employed to predict occupants’ count in each room. The models’ performances are evaluated using Root Mean Square Error (RMSE), Mean Square Error (MSE), Mean Absolute Error (MAE), and Normalized Root Mean Square Error (NRMSE). The proposed LGBM-XgBoost model outperforms other approaches for each type of space. Moreover, to highlight the importance of LGBM as a feature selection technique, the XgBoost model is also trained with all features. Results indicate that by selecting the appropriate features through LGBM, the RMSE and MAE for lecture rooms 1 and 2 are improved by 61.67%, 36.17% and 67.05%, 63.67%, respectively. Similarly, for office rooms 1 and 2 RMSE and MAE are improved by 33.37%, 71.5% and 59.7%, 51.45%, respectively.